NextFin News - Maggie Fanari, chief executive officer of J. Rothschild Capital Management and an investor in Anthropic, says the artificial-intelligence company can continue creating value after it goes public by compounding growth over the long term, with its focus on AI safety a strength rather than a constraint. The argument arrives as Anthropic weighs a public debut that could value the company at about $2 trillion — more than double the $965 billion post-money mark set by its $65 billion funding round in May.
The case rests on a pair of facts that point in opposite directions. Anthropic's revenue run-rate has raced from roughly $9 billion at the end of 2025 to more than $47 billion by May and $65 billion by the end of July, already ahead of rival OpenAI's reported $40 billion. In the second quarter of 2026 the company posted preliminary revenue of more than $11.5 billion, up from $787 million a year earlier, and swung to positive adjusted operating income for the first time. Yet its own prospectus shows a $42 billion net loss in 2025 and $518 billion of cloud, computing and infrastructure obligations still to come.
Fanari's thesis is ultimately a bet on which of those two facts the public market decides matters more.
The Situation: A Growth Story With a Capital Bill Attached
Anthropic is not arriving at the public market untested. The company confidentially filed a draft registration statement with the Securities and Exchange Commission on June 1, edging ahead of OpenAI in a closely watched race, and is working with Morgan Stanley, Goldman Sachs and JPMorgan on the sale. The debut, however, is now likely to be pushed to after the November US midterm elections, according to people familiar with the matter.
The growth trajectory is what underwrites the valuation. Revenue grew 12-fold in 2025 to nearly $4.6 billion, and the pace has accelerated since: the second quarter of 2026 alone produced more than $11.5 billion of revenue, versus $4.73 billion in the first quarter. The company has also projected 2028 revenue of roughly $190 billion to $200 billion, a figure that has become the anchor for the valuation discussions bankers are holding with investors.
The math behind the price tag is unforgiving. The $965 billion private mark equals roughly 15 times the $65 billion run-rate — rich by traditional software standards, but in line with where frontier labs have been valued in recent private rounds. The $2 trillion public target, by contrast, implies about 10 times the 2028 revenue projection. That is a lower multiple, but it is a multiple applied to revenue two years in the future, which is a longer look-ahead than public markets typically tolerate.
The profitability picture is split. On an adjusted operating basis, the second quarter of 2026 was positive — the company's first profitability signal. On a reported basis, 2025 closed with a net loss of nearly $42 billion, a figure that included roughly $34 billion of non-cash accounting charge tied to financing instruments that may eventually convert into shares rather than cash the company spent running the business. Stripped of writedowns linked to earlier fundraising, the 2025 operating loss was still more than $8 billion.
The capital requirement is the number that will define the IPO debate. Anthropic's prospectus lays out $518 billion of planned spending on cloud services and data centers over the coming years. That is not a discretionary budget line; it is the price of staying at the frontier. The company already spent $7.33 billion on compute and infrastructure last year, a threefold surge from 2024 that accounted for more than half of its $12.65 billion in total operating expenses. It held $20.28 billion in cash, cash equivalents and short-term investments as of December 31 — enough for a few years at the current burn, but not for the $518 billion plan.
There is also dependence risk. Nearly half of Anthropic's sales last year — 47% — were routed through cloud partners Amazon and Google, which are simultaneously its investors, its suppliers of computing power and its direct rivals in AI. Two unnamed customers each generated 12% of revenue last year, and the company warned that many of its largest clients are not locked into long-term contracts and could cut or stop spending.
Why Fanari's Thesis Is Not Just Optimism
Fanari's argument has two pillars: compounding growth and a safety advantage. Both have evidence behind them, and both are testable in public markets.
On growth, the mechanism is real. Anthropic's enterprise adoption has been driven by products such as Claude Code and Cowork, which embed the model inside workflows rather than confining it to a chat window. That shifts the revenue model from episodic API usage toward embedded, recurring enterprise deployment. Krishna Rao, Anthropic's chief financial officer, framed it plainly when the Series H round closed:
"Claude is increasingly indispensable to our growing global community of customers, and we work tirelessly to make tools like Claude Code and Cowork more helpful, more powerful, and more adaptable to their needs. This funding will help us serve the historic demand we are experiencing, stay at the research frontier, and bring Claude to more of the places where work happens."
On safety, the argument is that Anthropic's public-benefit-corporation structure and its willingness to advocate for regulation are not a drag on valuation but a durable moat. The company was founded by researchers who left OpenAI over governance and safety disagreements, and it has built its brand around the claim that capability and caution can advance together. In an environment where enterprise buyers are increasingly required to justify AI procurement to boards and regulators, a documented safety posture can be a purchasing criterion rather than a public-relations exercise. Fanari has made precisely this point: technological development and safeguards need to advance together, and Anthropic's safety focus is a strength rather than a constraint.
That positioning has already been stress-tested in public. The company clashed with the White House over the use of its tools, a dispute that led to the Pentagon temporarily blacklisting Anthropic — a move later blocked by a US judge. The episode shows both the cost and the credibility of the safety-first stance: it can cost government business in the short term, though government contracts account for less than 1% of annual revenue, while building a reputation that may pay off with regulated buyers in finance, healthcare and other compliance-heavy industries.
Brad Gerstner of Altimeter Capital, a co-lead of the Series H, put the investor read in blunt terms:
"Claude's latest advancements have driven large-scale adoption among the world's most demanding organizations. This momentum positions Anthropic to lead the next phase of AI innovation and capture the enormous opportunity ahead."
The Second-Order Question: Who Funds the $518 Billion?
The first-order read of Anthropic's numbers is that revenue is compounding faster than almost any company in history. The second-order question — the one the market has not fully priced — is not whether demand exists, but whether public investors will underwrite the capital intensity required to serve it.
Here is the transmission channel. Anthropic's growth depends on compute capacity, and compute capacity depends on data-center build-outs, chip supply and power — all capital-intensive, all long-lead. The $518 billion of planned infrastructure obligations means the company is effectively pre-committing to a spending path that must be financed. If public-market investors accept a $2 trillion valuation, Anthropic can raise equity and debt cheaply, lock in capacity, and keep compounding. If the multiple compresses toward conventional software norms, the cost of capital rises, capacity additions slow, and the growth rate that justified the valuation becomes harder to sustain.
This is why the IPO is more than a liquidity event for early investors. It is the moment the AI build-out shifts from private risk capital to public-market discipline. Marc Stad of Dragoneer, another Series H co-lead, captured the private-market mindset:
"The technological progress we are seeing right now is breathtaking. And we believe that we are still in the earliest days of both the development and commercialization of this technology."
Public markets, however, do not price earliest days. They price the next four quarters.
The structural implication cuts against both bulls and bears. For bulls: even if demand is real, growth can be throttled by the cost of capital. For bears: even if the multiple looks rich, a company that can finance $518 billion of capacity at a premium valuation can lock competitors out of the infrastructure they need to catch up. The winner may not be the lab with the best model. It may be the lab that can cheapest fund the data centers the model runs on.
There is a further twist inside the capital question: Anthropic's distribution runs through its competitors. Amazon and Google route 47% of its sales, and they also collected 60% of the $909 million in customer bills outstanding at the end of 2025. The prospectus itself acknowledges that this reliance "creates complex dynamics that could give rise to conflicts of interest and adversely affect our access to compute." The company frames the relationships as an advantage — access to the cloud platforms' vast sales networks — but the same channel gives Amazon and Google visibility into Anthropic's pricing and commercial terms, which could influence how aggressively they sell its products against their own models.
The Counter-Thesis: A Capital-Intensity Trap
The strongest case against Fanari's view does not question Anthropic's technology or its revenue trajectory. It attacks the unit economics underneath them.
The bear argument runs like this. A meaningful share of Anthropic's revenue is usage-metered inference spend that scales with compute costs rather than fixed-seat licensing, so gross margins are exposed to the same compute-price dynamics that drive the $518 billion capex plan. Customer concentration — two clients at 12% of revenue each, many without long-term contracts — means renewal risk sits in a small number of decisions. And the $42 billion reported loss, even with the $34 billion non-cash charge stripped out, shows the business is still burning cash on an operating basis. In this reading, Anthropic is not a recurring-revenue software compounder. It is a capital-intensive infrastructure business wearing a software multiple.
There is a second, sharper version of the bear case, and it comes from Anthropic's own rival. OpenAI has told investors and employees that Anthropic's accounting approach inflates its reported revenue by billions of dollars, because Anthropic books the full value of marketplace contracts — sales where customers buy access to Claude through a cloud provider's marketplace — as revenue, while recording the platforms' cut as a marketing cost. Anthropic has responded that it follows established accounting practices, recognizing gross revenue because it is the "principal" in the transaction. But if public-market analysts restate the numbers on a net basis, the growth rate and the multiple both come under pressure.
This is a serious argument, and it is the one the prospectus itself invites by foregrounding the loss, the spending obligations and the dependence on Amazon and Google for nearly half of sales. It is the default skepticism a public market applies to any pre-profitability listing at a nine-figure valuation.
The answer to it is equally concrete. Adjusted operating income turned positive in the second quarter of 2026 while revenue grew more than 14-fold year over year. That combination, sustained for even two more quarters, would show that operating leverage is real and that the loss is transitional rather than structural. The concentration risk is real but not unique: many enterprise-software franchises began with a handful of anchor clients before diversifying. And the safety moat, if it converts into procurement preference among regulated buyers, is exactly the kind of intangible asset that does not show up in the capex line but shows up in retention.
What Would Prove the Thesis Wrong
Fanari's case is falsifiable, and investors should watch three signals.
First, growth durability. If Anthropic's quarterly revenue growth falls below 40% year over year for two consecutive quarters after the IPO, the compounding story is broken and the multiple has no foundation. Second, margin trajectory. If adjusted operating income turns negative again after the second-quarter 2026 positive print, the operating-leverage argument fails and the capital-intensity bear case wins. Third, pricing. If the IPO prices below $1 trillion — half the $2 trillion target — the public market will have delivered its own verdict on whether private valuations can survive public scrutiny.
Conclusion: Value Creation Depends on the Cost of Capital, Not the Quality of the Model
The short-term read is volatility. A $2 trillion target on a company with a $42 billion reported loss will draw intense scrutiny, and the first public quarters will be judged against a bar no private round ever had to clear. The SpaceX precedent is instructive: the rocket company's June debut valued it at $1.77 trillion, its shares rose 19% on the first day to $160, but they have since pulled back to around $147 — above the $135 IPO price, but a reminder that debut enthusiasm and lasting value are not the same thing. The broader 2026 US IPO market has been the strongest since 2021, but strength in issuance volume has not prevented individual debuts from giving back first-day gains.
The medium-term read is fundamentals. If Anthropic can hold revenue growth above 40% while keeping adjusted operating income positive, the safety premium and the enterprise-embedded revenue model can justify a premium multiple. If it cannot, the $518 billion capex plan becomes a burden rather than a moat.
The long-term read is structural. The AI race is increasingly a race to fund infrastructure, and the public market is the cheapest permanent capital available. Anthropic's safety-first positioning may prove to be a durable advantage in enterprise procurement — but only if the company can access that capital at a valuation that does not mortgage its growth.
Three scenarios frame the path from here. In the base case, the IPO prices between $1 trillion and $1.5 trillion, growth holds above 40% for four quarters, and adjusted operating income stays positive — value creation continues, but the early public investors do not capture the full private-round premium. In the upside case, the company prices at or above $2 trillion, beats the 2028 revenue target ahead of schedule, and the safety moat converts into regulated-industry contracts; the stock compounds from the offer price. In the downside case, growth decelerates below 40%, adjusted operating income flips negative, and the multiple compresses toward single digits on revenue — the $518 billion build-out becomes a drag and value creation after the IPO turns negative for anyone who buys at the offer price.
Fanari is right that Anthropic can keep creating value after an IPO. But the mechanism is not the one most investors will be watching. They will be watching model benchmarks and revenue growth. What will actually decide the outcome is the cost of capital — and whether the public market is willing to fund a $518 billion build-out at a $2 trillion price tag.
Explore more exclusive insights at nextfin.ai.
